$ diff deepseek-v4 gemini-3-5-flash
DeepSeek V4 vs Gemini 3.5 Flash
DeepSeek and Google DeepMind, side by side. Every value below is the one recorded on the model’s own wiki page, with its citation — nothing is estimated.
DeepSeekDeepSeek V4
- context
- 1M
- weights
- open
- $/M in
- —
- $/M out
- —
- context
- 1.0M
- weights
- closed
- $/M in
- $1.5
- $/M out
- $9
What actually differs
- Context window
- Gemini 3.5 Flash takes 1.0M against 1M — modestly more room in a single request.
- Weights
- DeepSeek V4 publishes weights (MIT (open-weight)); Gemini 3.5 Flash is API-only. That decides self-hosting, air-gapped deployment and fine-tuning before any capability question does.
- Recency
- DeepSeek V4 shipped 58 days after Gemini 3.5 Flash (2026-07-16 vs 2026-05-19).
Full spec
| Attribute | DeepSeek V4 | Gemini 3.5 Flash |
|---|---|---|
| Developer | DeepSeek | Google DeepMind |
| Released | 2026-07-16 | 2026-05-19 |
| Context window | 1M tokens (384K max output) | 1,048,576 input / 65,536 output tokens |
| Pricing | Peak: 2× off-peak API rates; off-peak rates not disclosed | $1.50 in / $9.00 out per 1M tokens ($0.15 cached) |
| License | MIT (open-weight) | proprietary (API-only; no weight release) |
| Availability | DeepSeek API, Hugging Face (deepseek-ai/DeepSeek-V4-Pro) | Gemini API, Google AI Studio, Google Antigravity, Gemini app, AI Mode in Google Search |
Values come from DeepSeek V4 and Gemini 3.5 Flash, where each is cited to its source. This page states no benchmark result and ranks neither model — it puts two published specifications next to each other. Where a lab has not published a figure, the row says so rather than guessing. How these pages are produced.